kernel-hardening - Re: [RFC PATCH v3 1/8] Use refcount

Follow @Openwall on Twitter for new release announcements and other news

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CAPDLWs-fefTqAe+z-7BeALFpinanfPPd-9rmjKwUQ6WRP3_1Tg@mail.gmail.com>
Date: Tue, 19 Jan 2021 10:05:03 +0530
From: Kaiwan N Billimoria <kaiwan@...wantech.com>
To: Alexey Gladkov <gladkov.alexey@...il.com>
Cc: Linus Torvalds <torvalds@...ux-foundation.org>, LKML <linux-kernel@...r.kernel.org>, 
	io-uring <io-uring@...r.kernel.org>, 
	Kernel Hardening <kernel-hardening@...ts.openwall.com>, 
	Linux Containers <containers@...ts.linux-foundation.org>, Linux-MM <linux-mm@...ck.org>, 
	Andrew Morton <akpm@...ux-foundation.org>, 
	Christian Brauner <christian.brauner@...ntu.com>, "Eric W . Biederman" <ebiederm@...ssion.com>, 
	Jann Horn <jannh@...gle.com>, Jens Axboe <axboe@...nel.dk>, Kees Cook <keescook@...omium.org>, 
	Oleg Nesterov <oleg@...hat.com>
Subject: Re: [RFC PATCH v3 1/8] Use refcount_t for ucounts reference counting

(Sorry for the gmail client)
My 0.2, HTH:
a) AFAIK, refcount_inc() (and similar friends) don't return any value
b) they're designed to just WARN() if they saturate or if you're
attempting to increment the value 0 (as it's possibly a UAF bug)
c) refcount_inc_checked() is documented as "Similar to atomic_inc(),
but will saturate at UINT_MAX and WARN"
d) we should avoid using the __foo() when foo() 's present as far as
is sanely possible...

So is one expected to just fix things when they break? - as signalled
by the WARN firing?

--
Regards, kaiwan.


On Tue, Jan 19, 2021 at 2:26 AM Alexey Gladkov <gladkov.alexey@...il.com> wrote:
>
> On Mon, Jan 18, 2021 at 12:34:29PM -0800, Linus Torvalds wrote:
> > On Mon, Jan 18, 2021 at 11:46 AM Alexey Gladkov
> > <gladkov.alexey@...il.com> wrote:
> > >
> > > Sorry about that. I thought that this code is not needed when switching
> > > from int to refcount_t. I was wrong.
> >
> > Well, you _may_ be right. I personally didn't check how the return
> > value is used.
> >
> > I only reacted to "it certainly _may_ be used, and there is absolutely
> > no comment anywhere about why it wouldn't matter".
>
> I have not found examples where checked the overflow after calling
> refcount_inc/refcount_add.
>
> For example in kernel/fork.c:2298 :
>
>    current->signal->nr_threads++;
>    atomic_inc(&current->signal->live);
>    refcount_inc(&current->signal->sigcnt);
>
> $ semind search signal_struct.sigcnt
> def include/linux/sched/signal.h:83             refcount_t              sigcnt;
> m-- kernel/fork.c:723 put_signal_struct                 if (refcount_dec_and_test(&sig->sigcnt))
> m-- kernel/fork.c:1571 copy_signal              refcount_set(&sig->sigcnt, 1);
> m-- kernel/fork.c:2298 copy_process                             refcount_inc(&current->signal->sigcnt);
>
> It seems to me that the only way is to use __refcount_inc and then compare
> the old value with REFCOUNT_MAX
>
> Since I have not seen examples of such checks, I thought that this is
> acceptable. Sorry once again. I have not tried to hide these changes.
>
> --
> Rgrds, legion
>
>

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.